Article pubs.acs.org/JPCC
Low-Temperature Atomic Layer Deposition of Platinum Using (Methylcyclopentadienyl)trimethylplatinum and Ozone Jolien Dendooven,*,†,∥ Ranjith K. Ramachandran,†,∥ Kilian Devloo-Casier,† Geert Rampelberg,† Matthias Filez,‡ Hilde Poelman,‡ Guy B. Marin,‡ Emiliano Fonda,§ and Christophe Detavernier† †
Department of Solid State Sciences, COCOON, Ghent University, Krijgslaan 281/S1, 9000 Ghent, Belgium, Laboratory for Chemical Technology, Ghent University, Krijgslaan 281/S5, 9000 Ghent, Belgium, § Synchrotron SOLEIL, L′Orme des Merisiers, Saint-Aubin, BP48, 91192 Gif-sur-Yvette Cedex, France ‡
ABSTRACT: Thermal atomic layer deposition (ALD) of platinum is usually achieved using molecular oxygen as the reactant gas and deposition temperatures in the 250−300 °C range. In this work, crystalline thin films of metallic Pt have been grown by ALD at temperatures as low as 100 °C using (methylcyclopentadienyl)trimethylplatinum (MeCpPtMe3) as the Pt precursor and ozone as the reactant gas. The novel process is characterized by a constant growth rate of 0.45 Å per cycle within the 100−300 °C temperature window. The Pt films are uniform with low impurity levels and close-to-bulk resistivities even at the lowest deposition temperature. We show that the initial growth on SiO2 surfaces is nucleation-controlled and islandlike and demonstrate the good conformality of the low-temperature ALD process by Pt deposition on anodic alumina nanopores and mesoporous silica thin films.
■
an O2 plasma to grow Pt thin films at temperatures below 300 °C. Using the O2 plasma-based process, Longrie et al. obtained high purity Pt films in a broad temperature window from 150 to 300 °C.14 However, other work showed that the strong oxidizing power of the plasma can lead to oxidized platinum layers at temperatures below 250 °C.17,18 Therefore, Knoops et al. included a third reaction step in the ALD cycle to obtain high-purity Pt films at 100 °C: following the O2 plasma step, reduction of the oxidized platinum was achieved by exposure to H2 gas.17 Longrie et al. also reported ALD of Pt using the MeCpPtMe3 precursor and a N2 or NH3 plasma as reactant.14 Compared to the better known O2 plasma process, these processes are characterized by a faster nucleation on SiO2 substrates, resulting in very smooth Pt films. However, their temperature windows are more narrow and extend only from 250 to 300 °C for both reactants. In this work, we studied the growth kinetics, crystalline structure, resistivity, and purity of Pt thin films grown using O3 as reactant gas in combination with the MeCpPtMe3 precursor. It is shown that, despite the oxidizing power of O3, pure Pt thin films can be obtained even at a deposition temperature of 100 °C, making this process highly attractive for Pt deposition on thermally fragile substrates. Additionally, the conformality of the MeCpPtMe3/O3 process and its nucleation behavior on a SiO2 surface are discussed.
INTRODUCTION The self-limiting nature of the surface reactions in atomic layer deposition (ALD) enables the controlled deposition of ultrasmall amounts of platinum on complex 3D morphologies. This can be either as a thin conformal layer or, if growth is inhibited during nucleation, as dispersed nanoparticles.1−3 Given the excellent electric and catalytic properties of Pt, but also its high cost, ALD of Pt has therefore attracted considerable attention for applications in nanoelectronics, electrochemistry, catalysis, and sensing.4−9 The most commonly applied ALD process for Pt uses (methylcyclopentadienyl)trimethylplatinum (MeCpPtMe3) and O2 as precursors.10 The reaction mechanism of this process has been researched intensively in recent literature and relies on combustion-like reactions that occur during both ALD halfcycles.11−15 During the reactant pulse, O2 is dissociatively chemisorbed on the Pt surface, inducing combustion of the remaining ligands of the adsorbed MeCpPtMe3 molecules and the formation of a layer of adsorbed O atoms on the surface. These O atoms then react with some of the precursor ligands during the subsequent MeCpPtMe3 pulse. The optimum deposition temperature for this process is 300 °C.10,15 Higher temperatures cause thermal decomposition of the MeCpPtMe3 precursor, and for temperatures below 250 °C, very low growth rates are observed. Therefore, this process cannot be applied on heat-sensitive substrates. By using Pt(acetylacetonato)2 and O3 as precursors, metallic Pt thin films can be obtained at 140 °C.16 At 120 and 130 °C, this process results in amorphous platinum oxide thin films. Alternatively, the MeCpPtMe3 precursor can be combined with © 2013 American Chemical Society
Received: April 8, 2013 Revised: September 9, 2013 Published: September 11, 2013 20557
dx.doi.org/10.1021/jp403455a | J. Phys. Chem. C 2013, 117, 20557−20561
The Journal of Physical Chemistry C
■
Article
EXPERIMENTAL METHODS
All the depositions were performed in an experimental coldwall ALD chamber connected through a gate valve to a turbo pump backed up by a rotary pump. A second gate valve was installed for pre-evacuation of the chamber via a bypass line to the rotary pump. The solid MeCpPtMe3 precursor (99%, Strem Chemicals), kept in a stainless steel container, was heated above its melting point (30 °C), and the delivery line to the chamber was heated to 60 °C. Argon was used as a carrier gas for the Pt precursor. O3 was produced from a pure O2 flow with an OzoneLab OL100 ozone generator, resulting in an O 3 concentration of 175 μg/mL. During both ALD half-cycles a static exposure mode was applied, meaning that the valves to the pumping system were closed while exposing the sample to the precursor or the reactant gas. Unless stated otherwise, the pulse time of the MeCpPtMe3 precursor was 10 s, after which the valves to the pumping system were kept closed for another 5 s, resulting in a total exposure time of 15 s. For O3, the pulse time was 5 s and the total exposure time was 8 s. During the precursor and reactant exposures, the pressure in the chamber increased to ca. 5 × 10−1 mbar and 1 mbar, respectively. Between the two exposures, first the valve to the rotary pump was opened for 5 s, then the valve to the turbo pump was opened for another 45−50 s. In this way, the chamber was pumped down to a pressure of ca. 2 × 10−5 mbar. No purge gas was used. Film thicknesses were determined from X-ray reflectivity (XRR) and X-ray fluorescence (XRF) measurements. XRF was performed using a Bruker Artax system comprising a Mo X-ray source and an XFlash 5010 silicon drift detector. XRR and Xray diffraction (XRD) were carried out using a Bruker D8 system with Cu Kα radiation. In situ XRD measurements were acquired during annealing of the films in a home-built heating chamber mounted on the diffractometer.19,20 A linear Vantec detector with a range of 20° in 2θ was used to collect the diffracted X-rays at 3 s time intervals. The chemical composition of the deposited films was determined by X-ray photoelectron spectroscopy (XPS) using a Thermo VG Scientific ESCALAB 220i-XL with a monochromatic Al Kα X-ray source. Scanning electron microscopy (SEM) analysis was done on a FEI Quanta 200F instrument, and atomic force microscopy (AFM) was performed on a Bruker Dimension Edge system. In situ XRF measurements during ALD of Pt were performed on the SAMBA beamline at SOLEIL Synchrotron, France (proposal 20120293).21,22 The fluorescent radiation from the sample was captured with a Canberra 35elements planar germanium detector.
Figure 1. ALD temperature windows of the O3-based and the wellknown O2-based Pt processes. The O3-based process has a very broad temperature window ranging from 100 to 300 °C. The line serves as a guide to the eye.
correlated the low growth rates below 250 °C to a lower combustion rate of the precursor ligands during the O2 pulse.15 The present data suggest that O3 allows for more efficient removal of the precursor ligands without oxidizing the deposited Pt layer. Saturation of the ALD half-cycles was studied on sputtered Pt substrates at a deposition temperature of 150 °C. First, the pulse time of the MeCpPtMe3 precursor (followed by an additional 5 s static exposure) was varied. The experiments showed that a pulse time of 1 s, resulting in a pressure of 2 × 10−1 mbar in the chamber, was sufficient to reach saturated growth (Figure 2a). Similar experiments were then performed for the O3 step in the ALD process, indicating that saturation of the growth per cycle was achieved at an O3 pulse time of 3 s (followed by an additional 3 s static exposure) (Figure 2b).
■
RESULTS AND DISCUSSION The temperature window of the ALD process was investigated on Pt seed layers grown by sputter deposition to avoid nucleation problems. The substrates were placed on a heated sample stage, and the deposition temperature was varied from 60 to 350 °C. After the deposition of 60 ALD cycles, the thickness increase of the Pt layer was determined by XRR. A constant growth rate of 0.45 Å per cycle was obtained within a broad temperature window ranging from 100 to 300 °C (Figure 1). Nothing was deposited at 60 °C. At 350 °C, thermal decomposition of the MeCpPtMe3 precursor likely contributed to the growth, resulting in a slightly higher growth rate.10 Figure 1 also shows the temperature dependence of the growth rate for the well-known MeCpPtMe3/O2 process. Erkens et al.
Figure 2. ALD characteristics for the MeCpPtMe3/O3 process at 150 °C on sputtered Pt substrates. The dashed lines serve as guides to the eye. (a) Growth per cycle against the MeCpPtMe3 pulse time with a fixed total exposure time for O3 of 8 s. (b) Growth per cycle against the O3 pulse time with a fixed total exposure time for MeCpPtMe3 of 15 s. (c) Growth per cycle for various pump times used after the O3 step. 20558
dx.doi.org/10.1021/jp403455a | J. Phys. Chem. C 2013, 117, 20557−20561
The Journal of Physical Chemistry C
Article
1 lists the resistivity values measured on 15.5 nm thick Pt layers deposited on native SiO2/Si at different temperatures. With
Next, the effect of the pump time following the O3 step on the Pt growth rate was examined (Figure 2c). Increasing the pump time to several minutes resulted in a significantly lower growth per cycle, a behavior that can be explained based on the reaction mechanism of the Pt ALD process. Similar to the O2based and O2 plasma-based processes, a layer of adsorbed O atoms is expected to be formed on the Pt surface during the O3 exposure step. These O atoms are needed in the subsequent precursor pulse to react with the MeCpPtMe3 precursor molecules. When long pump times are used, however, a fraction of the O atoms may desorb, leading to a lower Pt growth per cycle. This effect was also observed for the NH3 and N2 plasmaenhanced ALD processes where unstable platinum nitride intermediates are likely formed on the surface during the plasma pulse.14 Linearity of the ALD process at 150 °C was verified on sputtered Pt and on native SiO2/Si substrates. The thickness of the deposited Pt films was extracted from XRF measurements. On the Pt surface, the film thickness increased linearly with the number of ALD cycles without nucleation delay (Figure 3a).
Table 1. Thickness, RMS Roughness, and Resistivity of Pt ALD Films Grown on Native SiO2/Si Substrates at Different Temperatures temperature (°C)
ALD cycles
thickness (nm)
RMS roughness (nm)
resistivity (μΩ cm)
100 150 300
300 250 200
15.8 15.5 15.3
0.3 0.3 0.9
21.2 19.3 18.9
150
400
24.0
0.3
11.9
increasing deposition temperature, fewer ALD cycles were needed to obtain a Pt thickness of ca. 15.5 nm, most likely because of a shorter incubation period at the start of the ALD process on the native SiO2 surface. The resistivity decreased slightly with increasing deposition temperature. The resistivity also decreased with increasing film thickness; a value of 11.9 μΩ cm (as compared to 10.8 μΩ cm for bulk Pt) was measured on a 24 nm thick Pt layer grown at 150 °C. The crystalline structure of the deposited Pt films was examined by XRD. Independent of the deposition temperature, the XRD pattern revealed (111) preferentially oriented Pt (Figure 4). Additionally, the effect of annealing was studied by
Figure 3. ALD characteristics for the MeCpPtMe3/O3 process at 150 °C on sputtered Pt and native SiO2/Si substrates. (a) Thickness of the deposited Pt films against the number of ALD cycles. The dashed line is a linear fit to the black square data points. (b) SEM image of a Pt layer deposited on a native SiO2/Si substrate using 100 ALD cycles.
On the native SiO2 surface, the growth rate was initially much lower than that on the Pt surface, but increased then to a higher value. This behavior is often indicative of island growth,23 a growth mode which was also observed for the thermal- and plasma-enhanced O2-based Pt ALD processes on metal oxide surfaces.24−26 SEM images confirmed the presence of islandlike Pt structures on the native SiO2 surface (Figure 3b). After 200 ALD cycles, the Pt layer deposited on the Si substrate appeared to be continuous in SEM and had a RMS roughness of less than 1 nm according to AFM measurements. The uniformity of the Pt layers was excellent, with a typical deviation in thickness of less than 2% across a two-inch area. XPS measurements were performed on Pt layers deposited at 100, 150, and 300 °C using 200 ALD cycles on a native SiO2/Si substrate. The C and O levels were below 5 atom % in all three Pt layers and the Pt4f7/2 peak position was 71.0 eV, indicative of metallic Pt.27 The deposition of metallic Pt at 100 °C is in contrast with the earlier reported O3-based process using the Pt(acetylacetonato)2 precursor, which resulted in PtOx below 140 °C.16 A possible reason for this different outcome at low temperatures could be the difference in reactor design, involving the use of higher process pressures in ref 16 compared to our work. The purity and metallic nature of the films was further confirmed by four-point probe resistivity measurements. Table
Figure 4. XRD patterns of a Pt ALD layer grown on Si at 150 °C and of the same Pt layer after annealing in He to 600 °C at a rate of 1 °C/s. The patterns were shifted for clarity. The inset shows the evolution of the XRD pattern (logarithmic scale) during this thermal treatment.
in situ XRD measurements.19,20 A 11.2 nm thick Pt film grown on native SiO2/Si at 150 °C was annealed in He to 600 °C at a heating rate of 1 °C/s. The inset of Figure 4 shows the evolution of the XRD pattern during this thermal treatment. The small (200) diffraction peak disappeared at 450 °C, while the (111) peak intensified. Moreover, additional peaks appeared symmetrically around the (111) diffraction peak. These peaks are the result of constructive interference of X-rays reflected from the surface of the Pt film and the interface with the Si substrate. Because a smooth surface is required to obtain interference fringes, the presence of these peaks suggests that the annealing induced the formation of smooth grains whose width is much larger than their thickness.28 20559
dx.doi.org/10.1021/jp403455a | J. Phys. Chem. C 2013, 117, 20557−20561
The Journal of Physical Chemistry C
Article
The conformality of the O3-based Pt ALD process was verified by deposition on a porous anodic alumina film on Si.29 The array of straight alumina pores with an average diameter of 23 nm was subjected to 250 ALD cycles at 150 °C using exposure times of 25 s for the MeCpPtMe3 precursor and 20 s for O3. Figure 5(a) shows a cross-sectional SEM image of the
Figure 6. XRF growth curves for Pt ALD at 150 °C in a mesoporous silica thin film and on a planar Si substrate. The inset shows the XRF growth curve for the planar substrate in more detail.
°C. The conformality of the process was proven to be excellent. We conclude that the low-temperature MeCpPtMe3/O3 ALD process is very attractive for applications that require the controlled deposition of Pt on heat-sensitive substrates, including polymeric surfaces.
Figure 5. (a) Cross-sectional SEM image of 23 nm wide anodic alumina nanopores coated with Pt. (b) Relative Pt coverage profile as obtained from EDX measurements. The Pt profile has been normalized to the average Pt coverage measured in the top part of the nanopores. The scan direction is indicated in (a).
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Phone: +3292648572.
coated pore structure. Pt was clearly deposited inside the nanopores down to a depth of ca. 0.8 μm, corresponding to a coated aspect ratio of ∼35. The depth profile of the Pt coating was quantified using EDX line scans and is shown in Figure 5(b). Conformal coating of larger aspect ratios should be possible by optimizing the deposition parameters. In view of catalytic applications where the support material often contains smaller mesopores (
